Phased array radio telescopes allow for the filtering of Radio Frequency Interference (RFI) in the spatial domain. Spatial filters are advantageous in radio astronomy when the separation between RFI and astronomical sources cannot be made in the time or frequency domains. Consequently, the mitigation of the RFI relies on the quality of its spatial signature vector (SSV) estimation. The latter depends on the array calibration information which is investigated in this work. More precisely, by using the Cramer-Rao Bound (CRB) tool, we evaluate the astronomical source power estimation error variance in presence of RFI for different array calibration scenarios corresponding to perfectly calibrated, direction-independent uncalibrated and direction-dependent uncalibrated array cases. In addition, we consider in this study the case where only the data covariance information is available and investigate the loss of performance due to the missing data with respect to different system parameters.